LinkedIn introduces ‘Seems like AI slop’ button for posts
▼ Summary
– LinkedIn added a “Seems like AI slop” button to its post menu, allowing users to flag content they believe a machine wrote, which then hides the post.
– The button uses the slang term “slop” for low-effort AI output, rather than neutral language like “AI-generated.”
– LinkedIn’s chief product officer confirmed the button helps train the platform’s classifiers to detect generic AI posts, following a May crackdown that suppressed such content from recommendations.
– Detection service Pangram estimated that 41% of long-form and 30% of short LinkedIn posts are likely AI-generated, highlighting the scale of the problem.
– LinkedIn’s approach is ironic because it offers its own AI writing assistant for Premium subscribers, while its parent company Microsoft is a major OpenAI backer.
LinkedIn has officially acknowledged what many users have long suspected: a significant portion of its feed is filled with low-effort, machine-generated content. The professional networking platform has introduced a new feature that lets users flag these posts with a blunt label: “Seems like AI slop.” This button, tucked under the three-dot menu on any post, signals a shift in how the platform is tackling the flood of generic, AI-written material.
When a user clicks the option, LinkedIn responds with a brief thank-you message and hides the post from their feed. According to 404 Media, which first reported the feature, finding a candidate for the button takes little effort. A quick scroll reveals posts marked by telltale signs: odd spacing, excessive emoji use, bullet-point takeaways, and the overused “X is not Y” phrasing. The choice of the word “slop” is intentional and striking. It avoids the clinical language of “AI-generated” and instead adopts a piece of internet slang that captures the low-quality, automated nature of the content.
The move follows LinkedIn’s broader crackdown in May, which aimed to suppress generic AI posts from recommendations without removing them entirely. Now, the company is turning detection into a crowd-sourced effort. Every flag helps train LinkedIn’s own models to better identify and filter out such content. Hari Srinivasan, LinkedIn’s chief product officer, confirmed the strategy in a post following the report. “AI slop is a top priority for all of us,” he wrote. The company is “ramping up a series of new and improved classifiers” and relying on member feedback to “tune our models and make better feeds.”
Srinivasan also added a crucial distinction: “AI and slop are not the same thing.” Many professionals use AI to refine their thoughts, so LinkedIn plans to test a private nudge in a poster’s analytics when their writing “may have come off as inauthentic.” Additionally, the platform is replacing its “enhance your post” feature with a proofreader that preserves the user’s original voice.
The scale of the problem is hard to ignore. Earlier this month, the detection service Pangram estimated that 41 percent of long-form LinkedIn posts and 30 percent of short ones are likely AI-generated. A quick search for “LinkedIn AI” reveals countless guides, many hosted on the platform itself, teaching users how to automate their posts. The irony is thick: LinkedIn offers its own AI writing assistant for Premium subscribers, while its parent company, Microsoft, is one of OpenAI’s biggest backers. The platform is effectively selling both the hose and the mop.
LinkedIn is not alone in this fight. The preprint server arXiv now bans researchers who submit AI-riddled papers, and other platforms have tightened rules on mass-produced AI channels. The pressure reflects a web where bots now outnumber humans in traffic terms. Whether a report button will truly move the needle remains uncertain. Slop is, as Srinivasan admitted, hard to define, and the definition keeps shifting. But asking users to name it, out loud, is a small admission that the feed has a problem. It also highlights a new reality: proving something is human is becoming its own kind of work.
(Source: The Next Web)
